Exercise, ergonovine and the cold pressor test have been used to provoke variant angina attacks. The sensitivity of these three tests was compared in 34 hospitalized patients with well documented, active variant angina who had recently undergone coronary arteriography. The three tests were usually performed on three consecutive days, and 28 of the 34 had the three tests within 1 week. Angina was provoked by ergonovine in all 34 patients, by exercise in 17 and by cold pressor test in only five (p less than 0.005). ST elevation developed during the ergonovine test in 32 (94%), during exercise in 10 (29%) and during the cold pressor test in only three (9%). With ergonovine, one patient had only ST depression and one had no ECG changes. During the cold pressor test two patients had pseudonormalization of abnormally negative T waves and 29 had no ECG changes. Exercise induced T-wave pseudonormalization in four patients, ST depression in nine others and no ECG changes in 11. ST elevation was more frequent with ergonovine than with either of the other tests (p less than 0.0001). ST elevation or T-wave pseudonormalization occurred more often with exercise than with cold (p less than 0.05), but both occurred less often than with ergonovine (p less than 0.0001). We conclude that the sensitivity of the ergonovine test is very high in patients with active variant angina and that exercise will provoke angina with ST elevation in about 30% of these cases. In contrast, the sensitivity of the cold pressor test is too low to be of much clinical value in the diagnosis of variant angina.